hope this helps someone who is using the Skywatcher EQ6-R and doing manual star alignmnet (3-Star-alignment).
Put up my rig last night as always for AP. I do polar alignment with the polar scope. I spent quite some time to get it really centered and I do level the tripod precisely which results in a polar error in the range of 3 arcmins, reproducable. Stacking in PI always shows field rotation less than 0.01° over hours, so I'm sure I'm doing well here. To my surprise the mount controller told me I'm 30arcmin off the pole. Convinced this is not true I slewed to the target and it was way of the image center. So I went back to calibration and this time picked stars that seemed more suitable to me, that is not too low at the horizon, not to close to each other and a long slew between eastern and western half of the sky to make sure cone error is calculated correctly. This time it reported 3 arcmin polar error, as expected. The mount and tripod was not touched and I'm sure it does not move (Berlebach Planet small, spikes sunk in the lawn with my full body weight - not gonna tell you how much that is).
If I recall correctly the controllers standard setting is sorting alignment stars by brightness, sorting by alphabet is also possible. Conclusion: it suggests the brightest stars, no matter if the position is mathematically problematic to build the model. It seems like one should not blindly use the suggested stars, so from now on I will pick them according to the above rules.
If anyone experienced something similar or knows details what the controller is doing, feedback is apprechiated!